4. COMPILATION OF DATA FOR THE CASE STUDY
4.4 Lake Kyoga long term lake levels and outflows
The hydrology of Lake Kyoga has been less extensively studied. The most in depth assessment of the components of its water balance has been published by Sutcliffe and Parks (1999), Sene (2000) and Brown and Sutcliffe (2013).
4.4.1 Lake Kyoga levels
Lake Kyoga levels have been monitored at Masindi Port, Bugondo and Kachung (Figure 4.5).
Figure 4.5 Long term lake levels and flow monitoring stations – Kyoga basin.
The record at Masindi port dates back from 1912 and Sutcliffe and Parks (1999) have derived a continuous record of lake level records from 10-day mean levels at Masindi Port for the
!
! . . !.
! .
! .
! .
! .
! .
! .
! .
Murchison Falls
Jinja Paraa
Fajao
Kibuze
Bugondo Kachung Kamdini
Mbulamuti Namasagali Masindi Port
Vict oria
Nile Kyoga Nile
Alb ert N
ile
Kafu
31
31
32
32
33
33
34
34
35
0 35 0
1 1
2 2
3 3
U G A N D A U G A N D A
K E N Y A K E N Y A
L. Kyoga
·
L. Albert
Legend
!
. Gauging stations
Rivers Lakes
International border Kyoga Basin
0 40 80 160
Kilometers 1:2,100,000
L. Victoria
period 1912-1977 as shown in Figure 4.6. This segment of the record is the most complete portion as later observations are riddled with gaps up until 1989.
Figure 4.6 Lake Kyoga monthly levels at Masindi Port 1912- 1977 (Sutcliffe and Parks, 1999).
Figure 4.7 illustrates the variation of concurrently recorded daily lake levels in metres above sea level (m.a.s.l) at Masindi port and Bugondo during the period 1948 – 2008. Prior to 1998 the two readings are shown to be closely correlated with minor differences arising due to application of different datums. A few sections of the missing record at Masindi port can be in- filled using observations made at Bugondo. However there is a brief common period where readings are not available at either of the two stations.
In May 1998, heavy rains dislodged floating masses of papyrus which blocked the outlet of Lake Kyoga, and modified the hydraulic gradient between the station at Masindi Port at the mouth and the station at Bugondo further upstream. The blockage was removed using mechanical equipment between the years 2000 – 2005 and thereafter the original relationship between Masindi Port and Bugondo gauge readings seems to have been restored.
Figure 4.7 Masindi Port and Bugondo gauge readings (1948 – 2008).
4.4.2 Lake Kyoga outflows
Along the Kyoga Nile, flow measurements have been made at Masindi Port, Kibuze, Kamdini, Fajao and Paraa. The station at Masindi Port is situated near the outlet of Lake Kyoga while the river gauge at Kamdini is located about 80 km downstream from Masindi Port. The station at Fajao is 1 km below Murchison Falls. Paraa is located 15 km from the point where the Kyoga Nile flows into Lake Albert (Figure 4.2). Gauging at Kamdini and Fajao, commenced in 1940.
Flows at Masindi Port from 1912 -1944 published in The Nile Basin (Hurst and Phillips, 1939;
Hurst and Black, 1945; Hurst et al., 1953) are based on a rating curve for Masindi Port derived from gaugings made at Fajao between 1907 and 1935. This curve was published by Hurst and Phillips (1938), and it was noted that the observations were fairly consistent over a long period.
The record before 1940 was based on few gaugings but these include measurements in 1922- 1923 when river levels were at their lowest. Thereafter, it was noted that the rating was not stable and flow measurements were discontinued (Hurst et al., 1946). From 1945-1951, the flow records are at Masindi Port were derived from a station at Kibuze located approximately
1030.0 1030.5 1031.0 1031.5 1032.0 1032.5 1033.0 1033.5 1034.0 1034.5 1035.0
01/01/1948 01/01/1968 01/01/1988 01/01/2008
Water level (m.a.s.l)
Date
Masindi Port Bugondo
40 km downstream of Masindi Port as published in the fifth supplement to volume IV of the Nile Basin (Hurst et al., 1957). The flow records at Kamdini for the period 1940-1955 published in The Nile Basin (Hurst and Phillips, 1939; Hurst and Black, 1945; Hurst et al., 1953, 1957, 1961), were based on annual gaugings. The record at Kamdini is considered to be more reliable than at Masindi Port, as the rating curve is more stable (Sutcliffe and Parks, 1999).
Since gaugings along the Kyoga Nile at the various locations have taken place intermittently and for varying periods of time, previous researchers have aggregated these measurements for purposes of compiling the most reliable flow along this reach at Kamdini after determining that tributary inflows along the Kyoga Nile are estimated to be less than 1% of the Victoria Nile flows and can reasonably be neglected (Sutcliffe, 1986; Gibb, 1989; Gibb, 1996; Sutcliffe and Parks, 1999; Sene, 2000). This aggregated record of Kyoga Nile flows has been taken to be synonymous with inflows to Lake Albert and outflows from Lake Kyoga. Figure 4.8 is an extract from Sutcliffe and Parks (1999) and it illustrates how the gaugings taken at Kamdini, Fajao and Para are plotted against the gauge readings at Kamdini to derive an aggregated record of Kyoga Nile outflows for the period 1940 to 1979.
Figure 4.8 Gaugings at Kamdini, Fajao and Paraa, 1940 – 1979 (Sutcliffe & Parks, 1989).
Sutcliffe and Parks (1999) compared the derived monthly outflows at Kamdini with Lake Victoria outflows for the common period 1940 – 1977. The method of deriving Lake Victoria outflows for the common period has been described in Section 4.3.2. Table 4.3 contains an extract from their publication and it summarises the main results of their regression analysis.
In Table 4.3, the tabulated parameters are defined after Sutcliffe and Parks (1999) as follows:
Coeff. is the regression coefficient b in an equation of the form Q0 = a + bQi, Seb is the standard error of b,
R, R2 are the coefficients of correlation and determination, See is the standard error of estimate, and
Constant is the Constant a in the equation.
Table 4.3 Regression of monthly outflows (m3 x 106 day-1) at Kamdini with Lake Victoria outflows, 1940-1977 (Sutcliffe & Parks, 1999)
Lag (t - months) Coeff. Seb R R2 See Constant
0 1.1319 0.0151 0.9620 0.9255 9.85 -8.03
1 1.1390 0.0139 0.9677 0.9365 9.09 -8.44
2 1.1336 0.0148 0.9633 0.9280 9.68 -7.92
3 1.1205 0.0169 0.9522 0.9067 11.02 -6.80
The results in Table 4.3 indicate that analysis with a lag of one month was the best fitting equation with an R2 value of 0.9365. Thus inflow to Lake Kyoga from the Victoria Nile (Qi) and outflow (Qo) based on the record at Kamdini can be estimated as:
Q0(t) = 1.139044 Qi (t -1) – 8.44181 (4.1) In order to apply Equation 4.1, monthly flows have to be converted to units of m3 x 106.day-1 in order to take account of the different month lengths. Assuming that the relationship in Equation 4.1 holds true during periods where outflow records from Lake Kyoga at Kamdini are not available, it can be applied to extend or infill the flow record for Lake Kyoga.
Gaugings along the Kyoga Nile resumed in 1996 and it is possible to deduce a new rating curve at Kamdini based on all available gaugings during the period July 1996 to June 2009 taken at Masindi Port, Kamdini and at Paraa, using the approach illustrated in Figure 4.6 as prescribed by Sutcliffe and Parks (1999). Two rating curves are discernible during this period as shown in Figure 4.9. There appears to be a significant shift in the rating in approximately 2000. The rating for the period March 2000 – June 2009 shows a wide scatter, indicating a large degree of uncertainty in estimation of the higher range of discharges. Changes in the rating curves in rivers without stabilised or lined channels are commonly observed in this region. They can be attributed to changes in cross section morphology of rivers due to the effects of sediment deposition, shifting papyrus movement and flooding (Petersen et al., 2007). In this instance the change in rating was taken into account in subsequent water balance calculations. The two rating curves can be adopted for purposes of extending the Kamdini record from 1996 to 2008.
Figure 4.9 Updated rating curves at Kamdini (1996 – 2009).
A complete monthly record of Kyoga outflows is required for the period 1899 – 2008 for purposes of subsequent application of the Equatorial Lake Model. This record was derived by concatenating five segments defined as follows:
y = 326.59x - 2731 R² = 0.8896 y = 419.15x - 3603.1
R² = 0.7638
600 700 800 900 1000 1100 1200 1300 1400 1500
10.60 10.80 11.00 11.20 11.40 11.60 11.80 12.00
Discharge (m3 .s-1 )
Kamdini gauge (m) February 1997 - February 2000 March 2000 - June 2009
(i) Application of Equation 4.1 to derive a representative record at Kamdini for the period 1899 – 1911 based on Lake Victoria outflows for the same period derived using the method outlined in Section 4.3.2.
(ii) Adoption of monthly outflows recorded at Masindi Port for the period 1912 – 1939 (Hurst and Phillips, 1939; Hurst and Black, 1945; Hurst et al., 1953).
(iii) Adoption of the monthly outflows recorded at Kamdini for the period 1940 – 1979 (Sutcliffe and Parks, 1999).
(iv) Application of Equation 4.1 to derive a representative record at Kamdini for the period 1980 – 1995 based on Lake Victoria outflows for the same period.
(v) Application of the rating curves illustrated in Figure 4.6 to derive outflows at Kamdini for the period 1996 – 2008.
Derivation of the long term Lake Kyoga outflows by concatenating different segments of data for purposes of analysing Lake Regulation was pioneered by Kennedy and Donkin (1996) and Mott MacDonald (1998).
Brown and Sutcliffe (2013) have assessed the suitability of utilizing the published early flows at Masindi Port (1915 – 1939) in a concatenation scheme designed to extend knowledge of Lake Kyoga outflows. The analysis utilized the insight gained from determining an updated annual water balance for Lake Kyoga for the period 1940 – 1977. The improved water budget estimates were realised to due quantification of net rainfall and tributary inflows. Thereafter Jinja annual outflow for the period 1915 – 1935 was adjusted based results of the improved water balance so as to obtain model derived estimates of Masindi Port outflow for the same period. To do so, the improved model was run with lake rainfall for the same period, evaporation based on a lake area of 4,700 km2 and local runoff estimated from basin rainfall.
The estimated flows were then compared with the published record at Masindi port (Figure 4.10).
Figure 4.10 Masindi Port: published and estimated annual flows, 1915 – 1939 (Brown and Sutcliffe, 2013).
Apart from the mismatch when the flood in 1917 is not predicted, the small difference between the estimated and published flows, suggests that the published flows at Masindi Port are a reasonable estimate of the true historic flows (Brown and Sutcliffe, 2013).
4.4.3 Extension of Lake Kyoga level records at Masindi Port
In order to compute the net basin supply time series for Lake Kyoga, through application of the Equatorial Lake model, a complete record of Lake Kyoga levels at Masindi Port for the period 1899 to 2008 was required. Beginning of month gauge levels at Masindi Port published in WMO (1982) for the years 1912 – 1947 were adjusted to end of month gauge levels. For example, the beginning of month gauge reading level for February 1912 was taken to be the end of month reading for January 1912. Thereafter, end of month gauge readings were extracted from the archives of the Directorate of Water Resources Management at Entebbe from the lake level station at Masindi Port so as to complete the record from 1947 up to the year 2008.
However, intermittent gaps still remained in this record from October 1978 – December 1997.
These missing levels were estimated in two steps.
In the first step, observations from lake gauging stations at Bugondo and Kachung were utilised after deducing the relationships amongst all lakeside gauges during periods of concurrent observation. To estimate the remaining portion of the record, an equation, derived by Mott MacDonald (1998) was adopted. This equation is essentially a fitted rating relationship to the
common period of mean monthly Masindi Port levels and monthly Kyoga outflows. The fitted equation is;
Qk = 0.00067352 * H5.717553 (4.2)
where, Qk = Lake Kyoga outflow in (m3.s-1), and H = Masindi Port mean monthly level (m).
Equation 4.2 estimates the lake level from available records of Kyoga outflow. It was also applied to estimate Lake Kyoga levels from 1899 to 1911 based on the representative record of Kyoga outflows derived using the approach described in Bullet (a) in Section 4.4.2.
Extension of the record of observed levels at Masindi Port from 1998 to the end of the year 2008 required a different approach. During the duration of the blockage, Masindi Port levels are not representative of the actual lake level. Koponen and Kummu (2004) noted that the net effect of the blockage was to cause an immediate rise of 1.5 m around the shoreline of Lake Kyoga and also a temporary decrease in the outflow. Koponen and Kummu (2004) also demonstrated that measured outflows along the Kyoga Nile did not correspond with the linear relationship that existed between water levels at Bugondo and Lake Kyoga yearly outflows prior to the blockage. Prevailing Lake Kyoga levels during the duration of the blockage, i.e.
from 1998 up to the end of the year 2003, have therefore been estimated using actual gauge readings at Bugondo recorded during the same period. After the blockage was effectively removed at the end of the year 2003, end of month Masindi port gauge readings for the period 2004 to 2008 were adopted.